skia /
skia /
93ed643df6d12cfaa553e0734e2828ef916a6b32 [graphite] Prefer 16-bit depth buffers
If there's no stencil attachment required, D16_unorm is pretty
universally available and half the memory of D32_float and since we
limit our depth uses to 16-bit precision anyways, it seems like the
preferred choice.
The combined depth/stencil logic is already preferring D24_S8 over
D32_S8 in Dawn and Vulkan, and I added the logic to use D24_S8 on
older Mac hardware. I believe there is some ongoing work needed in
Dawn itself to actually have its D24+_S8 format use D24 when
available on all backends.
Bug: b/341319155
Bug: b/306501391
Change-Id: I175773c403b908f8e6a702ff63e52e7aada411b7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/854377
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Nicolette Prevost <nicolettep@google.com>
15 files changed
